Documente Academic
Documente Profesional
Documente Cultură
Structural Behavior
of the Space Shuttle
SRM Tang-Clevis Joint
INASA-TH-89018) S'IRUCIURAL b E H A V I O B OF THE N87-11177
SPACE S H U T T L E S 6 f TANG-CLEVIS J C I H T :NASA)
32 P CSCL 20K
Unclas
G3/39 43511
September 1986
NASA
National Aeronautics and
Space Administration
Langley Research Center
Hampton, Virginia 23665
STRUCTURAL BEHAVIOR OF THE SPACE SHUTTLE
SRM TANG-CLEVIS JOINT
*
William H. Greene*, Norman F. Knight, Jr.t and Alan E. Stockwell*
SUMMARY
T h e space shuttle Challenger accident investigation focused on the failure of a tang-
clevis joint on the right solid rocket motor. T h e existence of relative motion between the
inner a r m of the clevis and the O-ring sealing surface on the tang has been identified as a
potential contributor to this failure. This motion can cause the O-rings t o become unseated
and therefore lose their sealing capability. Finite element structural analyses have been
performed t o predict both deflections and stresses in the joint under the primary, pressure
loading condition. These analyses have demonstrated the difficulty of accurately predicting
the structural behavior of the tang-clevis joint. Stresses in the vicinity of the connecting
pins, obtained from elastic analyses, considerably exceed the material yield allowables
indicating t h a t inelastic analyses are probably necessary. Two modifications have been
proposed t o control the relative motion between the inner clevis a r m and the tang at
the O-ring sealing surface. One modification, referred t o as the “capture feature”, uses
additional material on the inside of the tang t o restrict the motion of the inner clevis arm.
T h e other modification uses external stiffrning rings above and below the joint t o control
the local bending in the shell near the joint>. Both of these modifications are shown to be
effective in controlling the relative motion in the joint.
INTRODUCTION
T h e accident which destroyed the space shuttle Challenger is believed t o have been
caused by the failure of a case joint in the right solid rocket motor (SRM).’ A cross section
of this joint is shown in figure 1. T h e upper end of the lower cylindrical, motor segment
forms t h e clevis. T h e lower end of the upper cylindrical, motor segment forms the tang
which mates with the lower clevis. Around the circumference of both tang and clevis ends
are 180 holes into which one-inch-diameter connecting pins are inserted. Three of the
I n n e r clevis arm
pin holes on the tang end are used as alignment, slots to facilitate assembly of the SRM
segments. The seal between two motor segments is provided by two O-rings in the “inner
arm” of the clevis. The O-rings are compressed by a flat sealing surface on the tang.
Several characteristics of the original SRM joint design have been identified as po-
tential contributors to the failure. One characteristic is the behavior of the joint under
internal pressure load. The motor case expands radially outward due to the pressure. Be-
cause the joint has a higher hoop stiffness than the case wall on either side of the joint, its
radial expansion is less than that of the case wall. This nonuniform radial expansion is the
primary cause of relative motion between the inner clevis arm and the sealing surface on
the tang. This motion can cause the O-rings to become unseated and therefore lose their
sealing capability.
Following the accident, many studies have been undertaken, not only t o understand
this relative motion in the joint, but also to understand its overall structural behavior.
This paper reports the results of one such study carried out a t the NASA-Langley Re-
search Center. Specifically, this paper first considers the original SRM case joint. Finite
element analyses have been performed to predict both deflections and stresses in the joint
under the primary pressure loading condition. These deflections are compared with values
2
obtained from detailed tests performed at Mortoll-Thiokol.2 Following the development
and validation of t,he finite element model, two proposed modifications are considered in
this paper. One modification, referred to as the “capture feature”, uses additional mate-
rial on the inside of the tang to restrict the motion of the inner clevis arm. T h e other
inodificatiori uses external stiffening rings above and below the joint, to control the local
bending in the shell near the joint. Both of these modifications are designed to control the
relative motion between the inner clevis arm a n d the tang a t the O-ring sealing surface.
3
to construct the finite element mesh for each component. A typical finite element model
generated by this approach is shown in figure 2b.
T h e finite element system, EAL5, is used for all analyses reported in this paper. T h e
EAL language features along with the capability to easily add user-supplied computational
modules are exploited in the study. The finite element models for each component use the
six- and eight-node elastic, solid elements of the element library in EAL. These elements
are based on an assumed-stress hybrid formulation. The elements have elastic material
properties for all analyses performed in this study. A simple translation program is used
to convert from the SUPERTAB “ ~ n i v e r s a l ’description
~ of the model (e.g., joint locations,
element connectivities) to a format acceptable to EAL.
Joint C o m p o n e n t C o n n e c t i o n M o d e l i n g
A key ingredient of the joint modeling approach is the method used for connecting t h e
separate finite element models of tang, clevis, and pin. It was recognized from the outset
t h a t the contact regions between these components would change as a function of loading,
leading to a nonlinear analysis problem. This occurs, for example, when a gap t h a t exists
between unloaded components closes when loaded.
The actual contact is modeled by adding a nonlinear spring between two adjacent
contact nodes. The nonlinear spring stiffness curves are piecewise-linear functions of the
relative displacement and are generated, for example, such that a high stiffness results for
any compression of the spring and a low or zero stiffness results if the spring stretches. To
reduce the computational cost of the overall analysis, a substructure approach is adopted.
In this approach, all nonlinear behavior in the structure is assumed to be confined to specific
contact nodes on the boundaries of the components. Therefore, a reduced, nonlinear
problem can be formed which involves the equilibrium equations only at these contact
nodes.
This approach is implemented by using both EAL procedures and a n additional, new
computational module. The reduced stiffness matrices for each of the three components
are generated in EAT, using a “unit motion solution” technique. This technique involves
solving the linear system of equations for each case component once for each contact node
with a prescribed u n i t motion at that node. These sets of unit-motion solution vectors
(one set for each component) are used to transform the component stiffness matrices and
load vectors to corresponding reduced stiffness matrices and reduced load vectors. These
reduced component stiffness matrices along with the nonlinear spring stiffness definitions
are used by the new special-purpose module to form and solve the nonlinear equations.
4
ORIGINAL P!.Z I,
OF POOR QUALITY
-r
l - 7 .479
f
1.00
1
I I
All dimensions in inches.
4.15
1 T
1 2.808
1-
(a) Joint component dimensions.
TT
--.657
1.65
Pin
Tang
5
T h e reduced nonlinear equations are solved using a full Newton-Raphson technique a n d
convergence is rapid. T h e nonlinear solution vector of the overall problem is recovered
by using the solution of the reduced equations and the sets of unit-motion solution vec-
tors. This substructuring approach is considerably more convenient and cost effective t h a n
solving a set of non1inea.r equations for the overall system.
ORIGINAL JOINT
This section discusses the specific details of analyzing the original SRM tang-clevis
joint design. Deflections from analysis and tests are compared and analytical values of
stress are presented.
6
7':
I
+I G0011. 3111.324
L
"3 I
li
G002I. 2371.227
G0031. 2301.217
G0041. 2241.213
{: I II G0051. 2081.199
lkhie GOMI. 2031.200
GO071 2011 200
G0081: 2021: 202
G0091. 2071.211
G0101. 2111.218
GOll/. 217/. 237
' I
I
Ai.
/ I
1 G012I. 2241.312
I
I '
in the clevis arms and tang to be identical. Stiff springs are used to impose these frictional
constraints in the contact regions. The applied loading is 1000 psi internal pressure which
is assumed to act on the surfams of the inner clevis arm and tang down to the top of the
primary O-ring groove. The resulting axial component of this pressure load is also applied
and corresponds to a 36,500 lb/in. axial stress resultant in the shell wall (approximately
92,000 pounds of axial force at each pin).
A comparison of deflections from the referee test and analysis is shown in figure 3.
These deflections have been exaggerated in the deformed geometry plots. T h e agreement
between radial case deflections determined from strain gage measurements and analytical
values is generally good as indicated in figure 3a. The location of gage 12 is near t h e rings
used for attaching the external tank attachment rings. Although these rings are part of the
test specimen, the rings are not included in the present analytical model. T h e presence
7
- $ i " . ; - II
1
Relative displacement
between tang and
'
"
' 4
I
I
1 outer clevis a r m
I I
I
1
Re lat ive d isplace me nt
between tang and
i n n e r clevis a r m
(average gap motion 1
Radial/. 037/.0244
of the rings, however, explains the snialler radial deflections measured in this region as
compared with those predicted by analysis. Comparisons of the relative tang a n d clevis
motions in the joint itself as indicated in figure 3b are not as good as those comparisons
of overall radial deflections. T h e prediction of average gap motion between the tang a n d
the inner clevis arm midway between the two O-ring grooves, which is of major interest
in understanding the seal design, is approximately 34% below the average measured value
from the referee t,est. Possible causes for this difference are discussed in the following
sections.
8
(a) Model A (2737 nodes and 1516 elements).
Fig. 4 Three levels of discretization for the region near the pin.
first reason is to better account for pin flexibility. The second reason is to better model
the bending of the inner clevis arm. The close-up view of the deformed geometry shown in
figure 3b qualitatively indicates the severity of the pin bending due to the bearing forces
and t h e tang rotation. The model refinement from Model B shown in figure 4b to Model
C shown in figure 4c is motivated by a need for a better prediction of gap motion and
also better stress predictions in the vicinity of the pin. A main feature of Model C is two
additional rings of elements in the region where the pin contacts the inner clevis arm. These
additional elements allow better modeling of the contact force distribution. The pin itself
is also considerably more refined. Average gap motions obtained from analyses with Model
A and Model B were .0232 inches arid .0258 inches, respectively. The average gap motion
for Model C is .0244 inches and the corresponding deformed geometry plot is presented
in figure 3. From these results, we concluded that differences in gap openings between
test and analysis are not caused by poor finite element discretization. The prediction of
the structural behavior of the SRM tang-clevis joint must therefore be sensitive t o other
parameters associated with the joint itself and will be discussed in a later section.
9
(b) Model B (2953 nodes and 1676 elements).
Fig. 4 Continued.
10
I
il ~
Stress (ksi) = u /u
hoop axial
11
ksi). T h e main purpose of presenting these stresses is to demonstrate the need for analyses
which include plasticity effects and to indicate the areas where plastic deformations are
likely t o occur.
T h e outward bending of the tang has a significant effect on the stress distribution.
T h e compressive, contact stress a t the inside tang hole circle is over -300 ksi while at
the outside tang hole circle the stress is about -200 ksi. A tensile stress concentration
also occurs around the tang hole, approximately 90” from the maximum contact location.
Again, these high tensile stresses vary significantly from inside hole circle to outside hole
circle due t o tang bending.
At the edges of the pin holes in the tang and clevis, opposite the maximum contact
locations, there are stress concentrations due to the hoop loads. These tensile hoop stresses
are 209 ksi and 189 ksi at the lower edges of the clevis arm holes and 196 ksi and 170 ksi
at the upper edge of the tang hole.
There are particularly high compressive stresses a t the inner end of the pin in both
the hoop and axial directions. Because the yield stress of the pin material is substantially
higher than t h a t of the case material, the relief of these stresses in the actual hardware is
probably due mainly t o the yielding of the inner clevis arm.
Pin-Tang Clearance
Clearances between the three joint components (tang, clevis, pins) obviously must
exist t o satisfy assembly requirements. An analysis has been performed with a prescribed
uniform clearance around the pin circumference at the tang, and inner and outer clevis a r m
holes. Under the internal pressure load, the joint extended uniformly in the axial direction
until the clearance was closed. The joint behavior is essentially the same as without the
12
I nner clevis
Average
4
gap - clevis
mot ion, .02 arm
in.
-
I 1 I I 1 I
0.0 .002 .OM .006 .008 .010
clearance except t h a t the relative axial motion between tang and clevis increased by a n
amount equal t o the clearance. Similar behavior results if the clearance is prescribed only
a t the tang hole or only a t the two clevis a r m holes. Thus, this type of clearance does not
explain the discrepancy in gap motions.
A second clearance study is motivated by a careful scrutiny of the analytical stress
distributions in the vicinity of the pin. From figure 5, it can be seen t h a t the stresses in the
tang a r e larger a t the inside of the hole (nearest the shell centerline) t h a n a t the outside.
Tang bending causes both the compressive, contact stresses and the tensile stresses 90"
from t h e maximum pin contact location to be larger a t the inside hole circle than a t the
outside hole circle, Furthermore, these stresses considerably exceed the yield values for
the D6AC material. It is possible t h a t a differential clearance caused by material yielding
develops at, the tang holes. This differential clearance is larger a t the tang hole inside circle
t h a n at the outside circle (i.e., a tapered pin hole or1 the tang).
Results from a study with this type of clearance modeled are presented in figure 6. In
this study, a linear clearance or taper along the tang hole is assumed starting from zero a t
the outside hole circle to a value of A T at the inside hole circle. Figure 6 shows average
gap motion plotted as a function of AT. There is considerable change in gap motion as
13
A T is increased and for largc. A T , the averagca gap rriotion approaches the measured values
from tests.
.04
Average 1
gap
motion, 02
in. -1
-
Inner t
clevis Ac Outer
arm-’ clevis
arm
0.0 .001 .002 .003 .OM .OB
A,, In.
One of the specific purposes in refining the finite element mesh in the vicinity of the
pin (as discussed previously) is t o determine if this center-of-load location is sensitive t o
mesh refinement. Models A and B (see figures 4a and 4b) have only two nodes for contact
between pin and inner clevis arm along the pin centerline. T h e most refined model Model
C (figure 4c) has four nodes t o represent the contact force distribution in this region.
However, the gap motion predictions from these models are very close, indicating t h a t the
models represent, the pin-inner clevis arm contact distribution reasonably well for the ideal
joint.
14
However, this analytically-predicted contact force distribution may be incorrect for
the actual joint hardware. As mentioned previously, very high axial stresses are predicted
in the pin along its inside edge. The yielding that probably occurs in the inner clevis
a r m to relieve these pin stresses causes the center-of-load between pin and clevis t o move
outboard. A second possible effect on the contact distribution is how the pin end mates
with the inside of the inner clevis arm hole. The ideal finite element model assumes
perfect contact a t the inside of the hole. However, the chamfer at the pin end may make
this difficult to achieve resulting in less axial load being transmitted a t the pin end than
predicted by analysis.
T h e contact force distribution change caused by either of these effects can be studied
by specifying a small clearance between the pin and the inner clevis a r m in the finite
element model. This clearance is specified only a t the inner most circle of nodes on the
pin, and reduces the load transmitted between the pin and clevis arm a t this specific
location. T h e average gap motion as a function of this clearance is shown in figure 7. For
the larger values of clearance, average gap motions in the range of measured values for the
referee test (i.e., over .030 inches) are predicted.
Shim Thickness
Metal shims are inserted between the tang and outer clevis arm at each pin during
assembly to help control the clearance between tang and inner clevis arm a t the O-ring
sealing surface. Specified manufacturing tolerances on the tang allow the thickness a t the
sealing surface to vary between .777 and .792 inches. The inner spacing between clevis
arms varies between .827 and .842 inches. Thus, the total clearance at the O-ring location
can vary between .035 and .065 inches.
A study has been performed to determine the effect of shim thickness on gap motion.
Average gap motion is shown as a function of shim thickness for two cases in figure 8. T h e
first case specifies no clearance between the pin and tang a t the pin hole. The second case
specifies a pin hole taper equal to .007 inches. With no pin-tang clearance, the average gap
motion decreases only slightly with increasing shim thickness. This trend is inconsistent
with t h e results of reference 2 which report a measured reduction in gap opening of .008
inches when the shim thickness is changed from .034 to .050 inches. With .007 inches
of pin hole taper, there is a .004-inch reduct#ion in average gap motion when the shim
thickness is changed from .034 to .050 inches. These results reflect the trends reported for
t h e referee test and add credibility to the theory that tapering of the pin hole contributes
substantially t o gap motion.
15
c ’
.a0
.030
Average .020
gap A L
motion, T - f
in. A 1
T t
o Untapered pin hole
.OH
a Tapered pin hole
I 1
0
.030 ,040 .60 .060 .070
Shim thickness, in.
Pin Friction
Friction between the pins and tang and between the pins and clevis arms influences
the structural behavior of the tang-clevis joint. Clearances between the tang and clevis
arms allow the tang t o slide on the pins after assembly. This sliding is resisted only by
frictional forces bet,ween tang and pins or contact between the tang and the inner or outer
clevis arm.
T h e internal motor pressure increases from zero to near 1000 psi within 600 millisec-
onds after SRM ignition. During this pressure transient, the load on the pins changes from
compression due to the weight of the shuttle to tension due to the pressurization of the
motor. At some point in this transient, forces on the pins are theoretically zero and sliding
of the tang on the pins is possible. After this initial sliding, the axial load at each pin is
approximately 92,000 Ibs at SRM full pressurization which is probably sufficient to lock
the tang on the pins during flight. The analysis results presented in this paper assume this
locked condition.
To determine the effect of this assumption, an analysis has been performed where the
friction between tang and pin is neglected. This analysis assumes a total clearance between
tang and clevis arms of .065 inches. This clearance is partially taken up by a .034-inch shim
16
rl
@ Interference
between tang and outer clevis arm. The remaining .031 inches of clearance is assumed to
exist between tang and inner clevis arm. The analysis predicts an average gap motion of
-.019 inches (closing) for this case. This gap motion is the result of structural deform.ations
and the sliding of the tang on the pin toward the inner clevis arm. This result is significantly
different from the average gap motion of .024 inches (opening) presented previously for the
baseline analysis. Thus, it is clear that the assumption of friction between pin and tang is
significant in causing gap opening.
17
bend inward slightly and the inner clevis arm to bend outward (toward the tang) slightly.
T h e bending of the inner clevis arm towards the tang reduces the initial gap a t the O-ring
sealing surface.
As long as the magnitude of the interference is sufficient to insure zero clearance
between inner clevis arm and capture feature, the gap opening under the pressure loading is
essentially independent of the amount of interference. However, stresses will be introduced
in both tang and clevis components during assembly which add to the stresses resulting
from t h e external loads.
In the capture feature joint concept, it is possible to introduce a third O-ring in the
capture feature which seals against the inside surface of the inner clevis arm. An advantage
of this O-ring location is that, under increasing internal pressure, the O-ring becomes more
compressed. There are two main structural implications of including this third O-ring.
First, the O-ring groove removes a portion of the capture feature thus making it slightly
more flexible. Second, the pressure loading on the capture feature and inner clevis a r m
can be different from the case without, the third O-ring. With the third O-ring sealing, the
pressure is prevented from reaching the primary O-ring. In the studies reported herein,
without the third O-ring, the internal pressure acts on the capture feature and inner clevis
a r m u p t o the primary O-ring.
T h e finite element model shown in figure 9 is used in the analyses to determine the
structural behavior of the capture feature design. In this model, the capture feature has
a length of 1.603 inches and a thickness of .389 inches. Other dimensions of the joint are
given in figure 2a. An exaggerated deformed geometry plot of the modified joint with
capture feature is shown in figure 10 for the 1000 psi pressure load. T h e contact between
inner clevis arm and capture feature occurs at an axial location approximately equal to
t h a t of the bottom of the primary O-ring groove. For this case, the gap motion at the
bottom of t h e primary O-ring groove is .0051 inches and at the bottom of the secondary
O-ring groove is .0074 inches.
I Effect o f I n t e r f e r e n c e L o c a t i o n
T h e precise location where the interference between capture feature and inner clevis
a r m occurs can.be adjusted by tailoring the contact surfaces. It would be reasonable to
select a n interference point which results in optimum seal performance. A study has been
performed to determine the effect on gap motions of varying this interference location.
Figure 11 shows plots of gap motion distributions for four separate interference locations.
Both t h e gap motion measurements and interference locations are measured relative t o
18
L
the centerline between the two O-ring grooves. Case C results are shown in figure 10.
Case B, with the interference roughly opposite the bottom of the primary O-ring groove,
is optimum for reducing the average gap motion. However, interference Case C might be
a better alternative if the goal is to limit gap motion a t the top of the primary O-ring
groove.
19
Capture efaute
r,- T X
-012
.008
Gap
mot ion,
in.
.004
0
-1.0 0 1.0
Distance from centerline between O-ring grooves X, in.
performed. In this analysis, there are no external loads present, the tang is free t o slide
along the pin, and the maximum baseline interference is used. After “assembly”, the
average gap has closed approximately .026 inches indicating t h a t most of the deformation
occurs in the inner clevis arm rather than the capture feature. This result also indicates
t h a t improvements in the design from stiffening the capture feature will be small. T h e
interference force between capture feature and inner clevis a r m is 210 Ibs/degree. T h e
stresses induced due to the interference fit are small. The maximum stress is 12 ksi and it
occurs at the top edge of the pin hole in the inner clevis arm.
20
155/76\ ,,y- 152177
9
1991-204 86/17 -C
\
1741-143
-2321 -157
381209
-x/-
- -,
15216
-3501 -326 [-- -2061-166
411167 -160/-2722 j ,
-177110 165/ -300 -"T\
-- 1831-235
-821 -201
1f
!-1871-183
-2161 -203
Stress (ksi) = o Io
hoop axial
7 pq-
l 5 5 I 7 5 ?7-f 152178
introduced by the addition of the capture feature. As can be seen in figure 10, the capture
feature reduces the outward bending of t,he tang. As a result, the high stresses at the
inside of the joint around the pin are reduced relative to those on the outside of the joint.
For example, the maximum tensile, a.xial stress in the tang at the inside pin hole circle has
been reduced from 239 ksi to 209 ksi. Stresses in the inner end of the pin have also been
reduced considerably.
21
Graphite-epoxy composite is an attractive material system for the rings because of
its high extensional stiffness and low density. In addition, the motor can be manufactured
by filament winding the graphite fibers around the steel cases. A big drawback to the
graphite material is the difference between the coefficient of thermal expansion of the
graphite rings and the steel case. Graphite-epoxy composite has a coefficient of thermal
expansion approximately equal zero while the coefficient of thermal expansion for steel is
6.6 x lopG in./in./"F.
Many of the studies in this paper assume a baseline graphite ring design produced
by Morton-ThiokolG. This design uses a T40/982 graphite-epoxy system with Ehoop=
22.7 x loG psi. The coefficient of thermal expansion is assumed to be zero. Each ring is
7.5 inches long by 1.9 inches thick. The centerlines of the ring cross sections are located
f10.25 inches from the pin centerline.
22
control gap motiorl when added t o SRM case segments with
T h e rings can (~Ift~ctivcly
either the original joints or the rriodilied joints with the capture featiire. Both approaches
are consideixd iri t tic studics rcported herein. Ari cxaggerated deformed geornetry plot of
the capture feat urv design including tlic baseliiic~ring configuration is shown in figure 13.
T h e overall shell tleforinations shown i n figurci I:Ja arc considerably different from those
shown in figure I O a . The graphite rings have highcr hoop stiffness than the joint which
reverses the bending direction between t,hc rings and the joint. As a result, the average
gap between the tang and inner clevis arm closcs .013 inches under the 1000 psi internal
pressure load. Lkcaiise thv deformations are exagcratcd in figure 13b, this closing of .013
inches appears to cause the inner cltivis arm to overrun the tang. However, the initial
clearance betweeii the inl1t.r clevis a r m arid tlic tang is such that contact does not occur.
23
T
ll
O
-, r igi na I joint in.
Average \
gap
motion, n-. Modified joint with
in. capture feature
E = 22.7 x 106 psi
-.*01 O r
-.02 I I 1
0 1.0 2.0
Ring thickness t, in.
the original design because the capture feature itself adds more hoop stiffness at the joint
which must be counteracted by the ring hoop stiffness.
24
I
0,010
Ft
External
rings -("$
-Pin
Average 0.005 center Iine
gap I ,rMcdified joint with A
motion,
in.
0-
Original joint
-0.005 I 1
4.0 5.0 6.0 7.0 8.0 9.0 10.0
Location of rings from pin centerline LR, in.
because in these regions the capture feature is tending to limit the gap motion instead of
the relatively ineffective ring.
25
-Z
. .
I
8.5 x 10
w = radial deflection
oo= ring pretension stress
9 -
oO Ib 0
- 8.5 x I I 1 I I I I I
-15 -10 -5 0 5 10 15
allowable bondline stress values and allowable case distortions based on assembly require-
ments. T h e relationships between stresses, case dist,ortions, and temperature changes for
the case of the baseline rings attached to the capture feature design were established by
analysis. For a given value of ring pretension stress, un, and temperature change AT, t h e
radial case deflection under the ring, w , and hoop stress, o h o o p ,are given as
Note t h a t A T is measured relative t o the reference temperature where the ring pretension
stress c~ is prescribed. T h e bondline stress cannot be accurately predicted with the existing
finite element model. Therefore, it is calculated based on the hoop stress in the ring using
the familiar equation
where t and R are the ring thickness and radius, respectively. As a n example, if the ring
pretension stress is 2985 psi a t 7 0 " F , t'hen a t 20°F' ( A T = -50"), w = -.045 inches,
uhc,ol,= 0.0, and ubond = 0.0 are predicted using these equations.
26
For case assembly considerations, the deflection pattern at the case ends caused by t h e
ring pretension is also important. An analysis was performed with pretensioned, baseline-
design, rings on detached tang and clevis case segment ends. In figure 16, t h e radial
deflection normalized by ring pretension stress is shown as a function of distance from t h e
pin centerline. For reasonable values of pretension, such as the 2985 psi value considered
earlier, the unassembled case distortions are not large.
CONCLUDING REMARKS
Finite element structural analyses have been performed to predict both deflections
and stresses in the SRM tang-clevis joint under the primary pressure loading condition.
Correlations between test and analytical values of joint deflections have been made for
the original joint. The prediction of relative gap motion a t the O-ring sealing surface has
been emphasized because this motion is critical to the sealing capability of the joint. Two
modified tang-clevis joint designs have been evaluated.
T h e extensive analytical effort reported in this study indicates the difficulty in ac-
curately predicting the structural behavior of this tang-clevis joint. One contributor to
this difficulty is the problem of modeling the contact and clearances between the three
mating parts of the joint. Nonlinear analysis is required. A second contributor t o the
analysis difficulty is the high stress level in the vicinity of the pin for all joint compo-
nents. Compressive and tensile stresses are present which considerably exceed material
yield allowables. Analyses including inelastic material properties are certainly required to
accurately predict, these stresses.
27
I I
154179
156176
431 -1
3012
281 -1
2041 -220
$,+
1951 -107 - 2171 -173
-205/ -139
1521
‘’q-
16618
-126 \ g 1 9 8
145110
13217 14419
-3141 -295
[/ -2271 -184
c -148/-248’---r
/-T
-1, ,
-811 -184
-1331 -122 -2741 -266
Stress (ksi) = (I 1 (I
hoop axial
28
Several studies were performed to determine the sensitivity of the joint structural
behavior t o parameter changes. Two of these studies imposed clearance distributions
between pin a.nd tang, and pin and inner clevis arm that could be caused by material
yielding in these regions. The results show these clearances t o have a significant effect
on relative gap motion. This implies that analyses including inelastic material properties
may also be important in obtaining accurate predictions of joint deflections. A third study
varied the thickness of the metal shims between tang and outer clevis arm. Results from
this study show that increasing the shim thickness can slightly reduce gap motion at t h e
O-ring sealing surface.
Frictional forces between pin and tang have been shown to significantly affect gap
motion. When frictional forces are removed from the analysis, the gap closes rather than
opens.
Both the capture feature and external ring modifications to the original tang-clevis
design are effective in controling relative motion in the joint. The interference fit in the
capture feature design is critical to limiting the gap opening to between .005 and .007
inches. Residual stresses due t o the interference fit are low. External rings can be designed
which theoretically restrict gap motion to zero during pressurization. The effectiveness of
a given set of external rings depends on the ring hoop stiffness and location relative to
the joint. Studies have shown that the optimum ring locations are about seven inches on
either side of the joint. Ring hoop stiffness is a direct function of the material used. T h e
graphite-epoxy material considered in this study can be used in a practical design only if
problems associated with the coefficient of thermal expansion mismatch between the steel
case a n d the graphite-epoxy rings can be resolved.
29
REFERENCES
1. Report of the Presidential Commission on the Space Shuttle Challenger Accidend,
Washington, D.C., June 6 , 1986.
2. Oostyen, .J. E., Bright, D. D., Hawkins, G. F., McCluskey, P. M., and Larsen, G.
L., S R M Joint Deflection Referee Test: Phase 2 Ftnal Report, Morton Thiokol, Inc.,
Wasatch Operations, Document, Number TWH-300149, April 3,1986.
3. SDRC I-DEAS GEOMOD - Solid Modeling and Design Reference Manual, Structural
Dynamics Research Corporation, 1986.
6. SRM Case Field Jotnt and Case Nozzle I’resentatzon, Morton Thiokol, Inc., Wasatch
Operations, Document Number TWR-15623, Publication Number 86634, Presented
a t the Technical Interchange Meeting held at the NASA Marshall Space Flight Center,
Huntsville, AL, June 9-11, 1986.
30
1. Report No. 2. Government Accession No. 3. Recipient’s Catalog No.
NASA Thf-89018
4. Title and Subtitle 5. Report Date
Structural Behavior of the Space Shuttle SRM September 1986
Tang-Clevis Joint
6. Performing Organization Code
7. Author(s) 5&4b41-04
William H. Greene, Norman F. Knight, Jr. 8. Performing Organization Report No.
and Alan E. Stockwell
9. Performing Organization Name and Address 10. Work Unit No.
NASA Langley Research Center
Hampton, VA 236655225 11. Contract or Grant No.
For sale by the National Technical Information Service, Springfield, Virginia 22161
NASA-Lsnglcy, 1966